open-journal-matcher/csvdata.py

33 lines
986 B
Python
Raw Permalink Normal View History

2019-07-14 18:21:18 +00:00
""" extract issns from journallist.csv """
2019-03-17 22:47:24 +00:00
import csv
import json
import regex
2019-03-17 22:47:24 +00:00
2021-09-13 20:21:37 +00:00
MONTH = "2021-09"
2019-03-17 22:47:24 +00:00
output = []
2021-09-13 20:21:37 +00:00
with open("journallist-" + MONTH + ".csv", newline="") as csvfile:
2019-03-17 22:47:24 +00:00
data = csv.reader(csvfile)
for row in data:
try:
2021-04-18 04:33:58 +00:00
print(row[7].lower())
if "english" == row[7].lower():
if row[4]:
if regex.match(r'^[0-9]{4}-[0-9]{3}[0-9xX]$', row[4]):
output.append(row[4])
else:
print(row[4], 'regex does not match')
2021-04-18 04:33:58 +00:00
elif row[5]:
if regex.match(r'^[0-9]{4}-[0-9]{3}[0-9Xx]$', row[5]):
output.append(row[5])
else:
2021-04-18 04:33:58 +00:00
print(row[5], 'regex does not match')
else:
print("no issn")
except:
pass
2019-03-17 22:47:24 +00:00
2021-09-13 20:21:37 +00:00
with open("issnlist-" + MONTH + ".txt", "w") as issnfile:
2019-03-17 22:47:24 +00:00
issnfile.write(json.dumps(output))